Menu
Inshorts
For the best experience use inshorts app on your smartphone
inshortsinshorts
AI agent hacks McKinsey's internal AI platform in 2 hours in test, 46.5 mn chat messages exposed
short by Anmol Sharma / on Friday, 13 March, 2026
Security startup Codewall claimed its autonomous AI agent breached McKinsey's internal AI platform 'Lilli' in just two hours during a controlled test. The agent exploited an unauthenticated API endpoint and a SQL injection flaw to access the production database. The system exposed 46.5 million chat messages, 57,000 user accounts and 58,000 Word documents, among other data.
read more at Codewall